Optimal Self Tuning Neural Network Controller Design
| Authors: | Korosi Ladislav, Faculty of Electrical Engineering and Information Technology, Slovak University of Technology, Slovakia (Slovak Republic)
Kozak Stefan, Faculty of Electrical Engineering and Information Technology, Slovak University of Technology, Slovakia (Slovak Republic) |
|---|
| Topic: | 3.2 Cognition and Control ( AI, Fuzzy, Neuro, Evolut.Comp.) |
|---|
| Session: | Neural Networks in Modelling and Control |
|---|
| Keywords: | neural-network models, neural control, optimal control, on-line control, genetic algorithms |
|---|
Abstract
The proposed paper deals with modeling and control of continuous-time processes using artificial neural network with orthogonal activation functions, applicable for real-time control. A genetic algorithm has been used to find the optimal neural structure for on-line identification with the best learning algorithm. A moving prediction horizon in the control algorithm found by genetic algorithm has been compared with a constant prediction horizon. The proposed algorithms were verified on practical control problem and have proved a good performance.
